Pipelines Risk Assessment

THE INTERNATIONAL MARITIME TRANSPORT & LOGISTICS CONFERENCE (MARLOG 4)
A SUSTAINABLE DEVELOPMENT PERSPECTIVE FOR MEGA PROJECTS

29 – 31 MARCH 2015
Presentation Title: Risk Assessment of Cross

Country Pipelines using Fuzzy Clustering.
Name: Essam Mohamed Abdel Salam.

29 – 31 MARCH 2015
Outline:
Section Page
1. Introduction 3
2. Traditional indexing method 11
3. Fuzzy inference system 16
4. Subtractive clustering 24
5. Index sum & leak impact factor 25
database
6. Performance evaluation indices 27
7. Clustering the pipeline index sum & 28

leak impact factor data
8. Case study 34
9. Results & discussion 41
10. Conclusion 47
2
29 – 31 MARCH 2015
Introduction
• Pipelines are considered to be the safest,
cheapest, most efficient and reliable mean of
flammable substances transportation.
• Due to the huge volume of substances needed
to be transported, pipelines would be the only
possible mean for transporting the massive
quantities of petroleum.
3
29 – 31 MARCH 2015
Fig.1. Means of petroleum transportation

Pipelines move huge quantities of crude oil and refined products
efficiently, dependably and at a low cost, making many petroleum-based
products more affordable for Egypt consumers.
In addition, by reducing the volume of crude oil and refined products
that must be moved by truck or railway, pipelines make the roads and
railways safer for travelers.
4
29 – 31 MARCH 2015
• Although most of pipelines are located

underground and partially isolated from the
human interference but they are still
subjected to many threats and the product
leakage could lead to a harmful environmental
impact or human loss.
5
29 – 31 MARCH 2015
Fig.2. Pipeline leakage in Alberta Canada

Company not even aware of the leak until locals called to complain. Red
Deer, Alberta, Canada river watershed polluted.
6
29 – 31 MARCH 2015
• The pipe failure can never be fully avoided;

however, the overall risk of failure can be
reduced to an acceptable level by opting
efficient risk management strategies.
• Different risk assessment techniques are used
by O&G companies, including hazard and
operability (HAZOP) analysis, fault tree
analysis, scenario based analysis, and indexing
methods.
7
29 – 31 MARCH 2015
Fig.3. Risk assessment and the uncertainty problem.
8
29 – 31 MARCH 2015
• Pipeline risk assessment is complex and

imprecise due to the lack of information or
incomplete data. To deal with this uncertainty
in such situations, Fuzzy logic system
developed by Zadeh, (1965) can be used as a
decision making tool by processing linguistic
information of such complex structures where
this information is represented as fuzzy sets
inputs and the output risk values can be
represented as a crisp value or fuzzy sets with
associated degree of membership.
9
29 – 31 MARCH 2015
• In this study we employed the concept of fuzzy

logic in order to assess the risks of a pipeline.
• A number of models are established for the
Index Sum and the Leak Impact Factor of a
pipeline section.
• The performance of the constructed models is
evaluated in comparison with the hypothetical
calculated data and the best fit model is
identified based on the performance
evaluation indices, including Training RMSE,
Check RMSE, and correlation coefficient (R2). 10
29 – 31 MARCH 2015
2. Traditional indexing method

• A subjective scoring tool for assessing pipeline
risks based on a combination of statistical
failure data and operators experience where
the pipeline is sectioned to segments
according to factors of population, land type,
soil condition, coating condition, age of
pipeline or any other factors decided by the
evaluator.
11
29 – 31 MARCH 2015
Fig.4. Pipeline sectioning.
12
29 – 31 MARCH 2015
• Data is gathered from records and operators

interviews to establish an index for each
category of pipeline failure initiator, (a) third
party damage, (b) corrosion, (c) design, (d)
incorrect operations. These four indices score
the probability and importance of all factors
that increase or decrease the risk of a pipeline
failure. The indices are then summed to be
called Index Sum as shown in Eq. 1, as the
index sum score increases the probability of
risk decreases and vice versa.
13
29 – 31 MARCH 2015
• The last portion of the assessment addresses

the consequences of a pipeline system failure.
This consequence factor is called the leak
impact factor which is used to adjust the index
sum scores to reflect the consequences of
failure where a higher point represents a
higher risk.
LIF RRS
IS
IS
LIF RRS
14
29 – 31 MARCH 2015
• IS = TPD + C + D + IO (1)
• DF = LV / RE (2)
• LIF = LV × RE × DF × PH (3)
• RRS = IS / LIF (4)
Fig.5. The basic risk assessment model

15
29 – 31 MARCH 2015
3. Fuzzy inference system

• A Fuzzy Inference System (FIS) is a way of
mapping an input space to an output space
using fuzzy logic.
• FIS uses a collection of fuzzy membership
functions and rules.
• The rules in FIS (IF-THEN) are in the form:
− if p then q, where p and q are fuzzy statements.
For example, in a fuzzy rule
− if x is low and y is high then z is medium.
− Here x is low; y is high; z is medium are linguistic terms; x and y are
input variables; z is an output variable, low, high, and medium are
fuzzy sets.
16
29 – 31 MARCH 2015
• The functional operations in fuzzy inference

system proceed in the following steps.
− Fuzzification
− Fuzzy Inferencing (apply implication method)
− Aggregation of all outputs
− Defuzzification
Crisp input value

Fuzzification
Knowledge Inference
base engine
Crisp output value
Defuzzification
Fig.6. Fuzzy inference structure
17
29 – 31 MARCH 2015
• TS model introduced by (Takagi & Sugeno,

1985) is used to model complex non-linear
systems; its main feature is linearization of
each fuzzy rule as a linear subsystem. The
output is a blend of all these linear
subsystems which is done by aggregation of
rules.
18
29 – 31 MARCH 2015
• TS rules use functions of input variables as the

rule output (consequent). The general form of
TS rule model having two inputs x1 and x2, and
output U is as follows:
Where is a crisp function of the

output; are linguistic terms.
19
29 – 31 MARCH 2015
Fig.7. A typical inference mechanism for two input variables
20
29 – 31 MARCH 2015
• we have four variables for the Index Sum (IS)

model. Which are C, TPD, IO, and D.
The fuzzy IF-THEN rules of this model can be
defined as follows:
The parameters a, b, c, d, and e are estimated

from the training dataset of the IS model.
21
29 – 31 MARCH 2015
• And we have four variables for the Leak

Impact Factor (LIF) model. Which are PH, DF,
LV, and RE.
The fuzzy IF-THEN rules of this model can be
defined as follows:
The parameters f, g, h, i, and j are estimated

from the training dataset of the LIF model.
22
29 – 31 MARCH 2015
• The final output of the two fuzzy models is the

weighted average of all rule outputs in each
model, computed as:
• (6)
Where N is the number of rules, wi is the firing
strength to weight the ith fuzzy rule defined as:
• (7)
Where is the number of input variables;
is the grade of the membership function
23
29 – 31 MARCH 2015
4. Subtractive clustering
• Pipeline risk assessment by mathematical
modeling from expert's knowledge including the
input and output data of the system is possible.
• Fuzzy clustering technique is a powerful tool for
the identification of such system contains
possible uncertainty by grouping the input-
output data into fuzzy clusters then translating
these clusters into fuzzy IF-THEN rules, by this
we will avoid identifying all the rules as done in
conventional fuzzy inference method. 24
29 – 31 MARCH 2015
5. Index sum and leak impact factor

database
• Data of the (IS) and the (LIF) models are
obtained from experts opinions. Input
parameters of (IS) are TPD, C, D, and IO. While
input parameters of (LIF) are PH, LV, DF, and RE.
• The two models presented in this paper are
using a set of statical data consists of 625
input/output data, a part of this data is shown
in Table 1 for the (IS) model and Table 2 for the
(LIF) model. 25
29 – 31 MARCH 2015
Table.1. Statical description on dataset of IS model Table.2. Statical description on dataset of LIF model
26
29 – 31 MARCH 2015
6. Performance evaluation indices

• (8)
• (9)
• Where is the predicted values, is the the
qualitative expert's values, is the average of the
observed set, and is the number of data set.
• RMSE equals zero when the predicted output
exactly matches the recorded output.
• If equals zero it reflects a very bad correlation,
27
if equals one it reflects 100% correlation.
29 – 31 MARCH 2015
7. Clustering the pipeline index sum

and leak impact factor data
Table.3. Best selected model of IS
28
Table.4. Best selected model of LIF
29 – 31 MARCH 2015
Fig.5. IS fuzzy inference structure Fig.6. LIF fuzzy inference structure
29
29 – 31 MARCH 2015
Fig.7. Rules viewer and rules generated of IS model 30

29 – 31 MARCH 2015
Fig.8. Rules viewer of LIF model
31
29 – 31 MARCH 2015
Fig.9. Control surface of IS on (a) design & corrosion; (b) incorrect

operations & corrosion; (c) third party damage & corrosion 32
29 – 31 MARCH 2015
Fig.10. Control surface of LIF on (a) dispersion factor & product hazard;
(b) leak volume & product hazard; (c) receptors & product hazard 33
29 – 31 MARCH 2015
8. Case study
• Crude oil is discharged from the VLCC’s to the
SUMED pipeline via the SPM piping system.
• The tankers can then pass the Canal in ballast
condition with a low draft.
• Crude oil is dispatched through two parallel
pipelines, 42 inches diameter 320 km long,
starting from Ain Sukhna terminal to Sidi Kerir
terminal crossing the river Nile where a pressure
relief station preserves the pipeline from any
over pressure. 34
29 – 31 MARCH 2015
• After passing the canal in ballast condition,

the tankers are moored to a single point
mooring system (SPM) at the Sidi Kerir
terminal where the oil is reloaded to the
tanker via the terminal pumps and through
the SPM piping system.
35
29 – 31 MARCH 2015
Fig.11. Sumed pipeline sections
36
29 – 31 MARCH 2015
• The pipeline has different wall thicknesses

varied from 11.13 mm to 22.22 mm according
to design.
• The 320 km pipeline will be sectioned to 7
sections with varied distances.
• Sectioning the pipeline is done by putting the
following phenomenon in consideration; the
type of land, soil condition, atmospheric type,
population density, Crossing River & water
ways, high/low lands, and finally the existence
of Right of Way (ROW). 37
29 – 31 MARCH 2015
Table.4. Pipeline sections & characteristics
• The risk assessment is performed by using the

traditional method and by the proposed
model on each pipeline section separately.
38
29 – 31 MARCH 2015
• The section pipeline with the lowest RRS value

is selected as the riskiest section, that may
help the pipeline operators to start managing
the risk on the lowest score pipeline section.
• In the lowest RRS value section the operator
may start with the lowest scored index to
improve the reliability and safety of this
section, e.g. low scored design index.
39
29 – 31 MARCH 2015
• The results of the traditional RRS method for

risk assessment of 7 sections are calculated,
based on Eq. 1, 2, 3, and 4. An example is
presented as follows:
• IS(section1) = 84 + 83 + 1 + 82 = 250
• DF(section1) = 2/2 = 1
• LIF(section1) = 9×2×1×2 = 36
• RRS(section1) = = 250/36 = 6.94
40
29 – 31 MARCH 2015
9. Results and discussions
Table.5. Output RRS results of the proposed model
• Section number 3 of the pipeline has the

lowest RRS value and ranked as the riskiest
part of the pipeline as it passes through the
river Nile. 41
29 – 31 MARCH 2015
• Section number 3 will be the starting point in risk

management to decrease the risks on it.
• The risk assessor can start by enhancing the
design index record of this section as it has the
lowest value between the index sum indices.
• The design index record can be enhanced by
doing the following: Increase Pipe Safety Factor,
Increase System Safety Factor, Avoid Fatigue,
Avoid Surge Potential, Make a System Hydrotest
to ensure pipeline integrity, Avoid Pipe
Movements. 42
29 – 31 MARCH 2015
• A comparison between traditional method

and proposed model shows very high
correlation and proves that the proposed
model using fuzzy clustering is a powerful tool
for pipe line risk assessment.
Table.6. Output RRS results of traditional method and proposed model
43
29 – 31 MARCH 2015
Fig.12. RRS results of traditional method versus proposed model
44
29 – 31 MARCH 2015
• As depicted in Fig. 13. The high correlation

coefficient value (0.9999) for index sum,
(0.9903) for leak impact factor, and (0.9821)
for RRS, implies the effectiveness of using the
TS fuzzy inference method based on
subtractive clustering.
45
29 – 31 MARCH 2015
Fig.13. Correlation coefficient degree between qualitative and subtractive clustering model
46
29 – 31 MARCH 2015
10. Conclusion
• In this paper the indexing pipeline risk
assessment methodology is integrated with
subtractive clustering fuzzy logic to deal with
the uncertainty of the real world conditions
and to avoid the difficulties of constructing
many rules as the computational complexity
increases with the dimensions of the system
variables because the number of rules
increases exponentially as the number of
system variables increases. 47
29 – 31 MARCH 2015
• The pipeline is divided to seven sections and

the risk assessment procedure is done for
each section by both qualitative and proposed
model.
• The results showed that computed RRS values
using proposed model are consistent with
those obtained using qualitative method. The
results also showed a high correlation and
high accuracy of the proposed model.
48

Pipelines Risk Assessment

Uploaded by

Copyright:

Available Formats

You might also like

Pipelines Risk Assessment

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Pipelines Risk Assessment

Uploaded by

Copyright:

Available Formats

THE INTERNATIONAL MARITIME TRANSPORT & LOGISTICS CONFERENCE (MARLOG 4)

A SUSTAINABLE DEVELOPMENT PERSPECTIVE FOR MEGA PROJECTS

Presentation Title: Risk Assessment of Cross

Name: Essam Mohamed Abdel Salam.

7. Clustering the pipeline index sum & 28

Fig.1. Means of petroleum transportation

• Although most of pipelines are located

Fig.2. Pipeline leakage in Alberta Canada

• The pipe failure can never be fully avoided;

Fig.3. Risk assessment and the uncertainty problem.

• Pipeline risk assessment is complex and

• In this study we employed the concept of fuzzy

2. Traditional indexing method

Fig.4. Pipeline sectioning.

• Data is gathered from records and operators

• The last portion of the assessment addresses

Fig.5. The basic risk assessment model

3. Fuzzy inference system

• The functional operations in fuzzy inference

Crisp input value

Fig.6. Fuzzy inference structure

• TS model introduced by (Takagi & Sugeno,

• TS rules use functions of input variables as the

Where is a crisp function of the

Fig.7. A typical inference mechanism for two input variables

• we have four variables for the Index Sum (IS)

The parameters a, b, c, d, and e are estimated

• And we have four variables for the Leak

The parameters f, g, h, i, and j are estimated

• The final output of the two fuzzy models is the

5. Index sum and leak impact factor

6. Performance evaluation indices

7. Clustering the pipeline index sum

Table.3. Best selected model of IS

Fig.5. IS fuzzy inference structure Fig.6. LIF fuzzy inference structure

Fig.7. Rules viewer and rules generated of IS model 30

Fig.8. Rules viewer of LIF model

Fig.9. Control surface of IS on (a) design & corrosion; (b) incorrect

• After passing the canal in ballast condition,

Fig.11. Sumed pipeline sections

• The pipeline has different wall thicknesses

Table.4. Pipeline sections & characteristics

• The risk assessment is performed by using the

• The section pipeline with the lowest RRS value

• The results of the traditional RRS method for

9. Results and discussions

Table.5. Output RRS results of the proposed model

• Section number 3 of the pipeline has the

• Section number 3 will be the starting point in risk

• A comparison between traditional method

Table.6. Output RRS results of traditional method and proposed model

Fig.12. RRS results of traditional method versus proposed model

• As depicted in Fig. 13. The high correlation

• The pipeline is divided to seven sections and

You might also like